Apple Intelligence is designed to protect your privacy at every step. It’s integrated into the core of your iPhone, iPad, and Mac through on-device processing. So it’s aware of your personal information without collecting your personal information. And with groundbreaking Private Cloud Compute, Apple Intelligence can draw on larger server-based models, running on Apple silicon, to handle more complex requests for you while protecting your privacy.
https://www.apple.com/apple-intelligence/
Apple Intelligence

Apple Intelligence is for the everyday and it‘s deeply integrated into iPhone, iPad, Mac, and Apple Vision Pro with groundbreaking privacy.

Apple

🆕 on the Security Blog: Private Cloud Compute

Built with custom Apple silicon and a hardened operating system, Private Cloud Compute extends the industry-leading security and privacy of Apple devices into the cloud, making sure that personal user data sent to PCC isn’t accessible to anyone other than the user — not even to Apple.

https://security.apple.com/blog/private-cloud-compute/

Private Cloud Compute: A new frontier for AI privacy in the cloud - Apple Security Research

Secure and private AI processing in the cloud poses a formidable new challenge. To support advanced features of Apple Intelligence with larger foundation models, we created Private Cloud Compute (PCC), a groundbreaking cloud intelligence system designed specifically for private AI processing. Built with custom Apple silicon and a hardened operating system, Private Cloud Compute extends the industry-leading security and privacy of Apple devices into the cloud, making sure that personal user data sent to PCC isn’t accessible to anyone other than the user — not even to Apple. We believe Private Cloud Compute is the most advanced security architecture ever deployed for cloud AI compute at scale.

Private Cloud Compute: A new frontier for AI privacy in the cloud - Apple Security Research

Apple Intelligence orchestrates between on-device models (eg. summarization, tone adjustment, refining ... ) and inference on the Private Cloud Compute infrastructure.

Find out more about these Apple-developed models in the Platform State of the Union. https://developer.apple.com/wwdc24/102?time=115

Platforms State of the Union - WWDC24 - Videos - Apple Developer

Discover the newest advancements on Apple platforms.

Apple Developer

Introducing Apple’s On-Device and Server Foundation Models

Learn more about the 3B parameter on-device language model and the larger server-based language model available with Private Cloud Compute.

https://machinelearning.apple.com/research/introducing-apple-foundation-models

Introducing Apple’s On-Device and Server Foundation Models

At the 2024 Worldwide Developers Conference, we introduced Apple Intelligence, a personal intelligence system integrated deeply into iOS 18…

Apple Machine Learning Research
Apple Intelligence Foundation Language Models

We present foundation language models developed to power Apple Intelligence features, including a ∼3 billion parameter model designed to run…

Apple Machine Learning Research
"Creating the trust model where your device will refuse to issue a request to a server unless the signature of all the software the server is running has been published to a transparency log was certainly one of the most unique elements of the solution—and totally critical to the trust model"
https://infosec.exchange/@mattburgess/113124728485246481
Matt Burgess (@[email protected])

Apple's new AI iPhones are underpinned by some seemingly clever privacy engineering and protections. @[email protected] spoke with Craig Federighi about how it all works https://www.wired.com/story/apple-private-cloud-compute-ai/

Infosec Exchange
@fj I see that Apple, just like everyone else, is using publicly available data unless someone opts out. I wish Apple had done more to curate the training dataset and better protect people’s intellectual property, not just do the bare minimum.

🆕 The Private Cloud Compute Security Guide covers a variety of topics including our single-use unlinkable authentication tokens and the append-only and cryptographically tamper-proof transparency log ensuring non-targetability

https://security.apple.com/documentation/private-cloud-compute

https://hachyderm.io/@lukasa/113363594603927141

Documentation

@fj breaking that is sooo tempting...

@fj That must have been really fun to come up with! It’s a really cool take on the traditional trust model of public cloud operators. I think that the public enforcement angle is a really good idea. Failing to properly maintain the transparency log will hurt Apple enough that you will make sure not to mess up.

OS updates for e.g. iOS do not (yet) jump through those public audibility hoops, right? Therefore, pushing a malicious OS update that breaks the client-side trust is still possible?

@fj I'm still not sure how this is going to work.

It doesn't matter how private is when working in normal operations, if you can just 'steer' user devices to honeypot servers, it's useless. (hence why all the 'you don't have to trust apple' marketing speak around iCloud Private Relay is a complete fabrication.)

They do mention Intel SGX & AWS Nitro, which is likely going to be more private than Apple PCC, because with those, the entity doing the attestation is independent from the entity that controls the servers (i.e. Microsoft software running on an Intel CPU), but with Apple, it's Apple software running on Apple Silicon.